Scalability Analysis of KVM-Based Private Cloud For IaaS

Abstract

One of the cloud technology cores is virtualization. Virtual Machine Manager (VMM) is said to have good scalability if it provides services to many virtual machines with a fair management of resources to maintain optimal performance of virtual machines’. Scalability evaluation of virtualization technology needs to be done so that cloud developers can choose the appropriate VMM according to the scenario of cloud usage. This study was conducted to determine the scalability of KVM in a cloud with OpenStack platform. Three scalability metrics were used (overhead, linearity, and isolation) to measure the scalability of different machine resources: CPU, network, and disk. The results showed that KVM exhibits good scalability in CPU and network. KVM is suitable for a scenario in which isolation between its CPU and harddisk is needed. KVM is suggested not to be used in a scenario where harddisk is accessed by many VMs intensively.